Field of the Invention
The present invention relates to an image reading device that includes a duplex parallel reading function and a non-transitory readable recording medium. The present invention more specifically relates to a technique of detecting a specific document.
Description of the Background Art
An image reading device that includes a function to read both sides of a document at the same time has an image reading unit for an obverse side and a back side of the document at a separate place from each other in a document path. Therefore, an image reading process is performed for each of the obverse and back side read at the image reading device. More specifically, in most cases, each processing unit that processes image data generated by reading the document is required for the obverse side and the back side of the document, respectively. Thus, two processing units are required for the entire device.
Some of the image reading devices have a function to determine whether or not the read document is one of specific documents such as a bank bill or securities. Some of the image reading devices are capable of restricting a forgery of the specific documents by terminating the reading process of the document when determining that the read document is one of the specific documents. The image reading device provided with the processing units for each of the obverse and back side of the document generally includes two processing units for each of the obverse and back side of the document for the determination process of the specific documents. In such a case, the number of the processing units necessary for the image reading device becomes large. A circuit size of the device becomes twice as large, and a structure of the device becomes complicated. Also, it is expensive to manufacture the devices.
An image processing device is known which is capable of reducing each image data of the obverse side and the back side of the read document in main scanning direction, and creating a linked image by lining and linking the two reduced image data. The image processing device then determines whether or not the linked image is the specific document. This known technique is introduced for example in Japanese Patent Application Laid-Open No. JP 2005-26880 A. According to the image processing device, only one processing unit that determines whether or not the image data is the specific document is enough, and two are unnecessary. The image processing device according to the known technique, however, has also some problems described below. FIGS. 16A, 16B and 16C show an example when obverse side image and back side image of the document are linked according to the known technique. FIG. 16A shows an example of characteristic patterns 101 and 102 for specific document determination that are embedded in the obverse side of the document. As shown in FIG. 16A, for example, the two characteristic patterns 101 and 102 are embedded in the obverse side of the document. A process to divide a whole image Ga into multiple blocks B100 corresponding to the size of the characteristic patterns 101 and 102 and determine one by one whether or not each block B100 includes an image component that forms at least a part of the characteristic patterns 101 and 102 is performed to detect the characteristic patterns 101 and 102 in the image Ga created by reading the obverse side of the document. FIG. 16B shows an example of a characteristic pattern 103 for the specific document determination that is embedded in the back side of the document, in an image Gb of the back side of the document. The images Ga and Gb of the obverse side and the back side of the document read by a reading unit, respectively, are reduced one by one and linked to each other as shown in FIG. 16C, then a single image Gc is created. One processing unit is required for the determination whether or not the created image Gc is the specific document.
In the known technique, when the two image data are compressed into half of their original size, respectively, the number of the blocks B100 that should be read for the detection of the characteristic patterns 101, 102 and 103 is no different from the two images of the obverse side and the back side. Then, the number of the blocks that should be read for creating the image Gc is still two, which is not changed. For performing the process to detect the characteristic pattern by the single processing unit, double the time is required as compared for the process performed for the single image.